// Name: common/sstream.cpp
// Purpose: string-based streams implementation
// Author: Vadim Zeitlin
-// Modified by:
+// Modified by: Ryan Norton (UTF8 UNICODE)
// Created: 2004-09-19
// RCS-ID: $Id$
// Copyright: (c) 2004 Vadim Zeitlin <vadim@wxwindows.org>
// ============================================================================
// ----------------------------------------------------------------------------
-// seek/tell
+// construction/destruction
+// ----------------------------------------------------------------------------
+
+// TODO: Do we want to include the null char in the stream? If so then
+// just add +1 to m_len in the ctor
+wxStringInputStream::wxStringInputStream(const wxString& s)
+#if wxUSE_UNICODE
+ // FIXME-UTF8: use wxCharBufferWithLength if we have it
+ : m_str(s), m_buf(s.utf8_str()), m_len(strlen(m_buf))
+#else
+ : m_str(s), m_buf(s.mb_str()), m_len(s.length())
+#endif
+{
+#if wxUSE_UNICODE
+ wxASSERT_MSG(m_buf.data() != NULL, wxT("Could not convert string to UTF8!"));
+#endif
+ m_pos = 0;
+}
+
+// ----------------------------------------------------------------------------
+// getlength
// ----------------------------------------------------------------------------
-off_t wxStringInputStream::OnSysSeek(off_t ofs, wxSeekMode mode)
+wxFileOffset wxStringInputStream::GetLength() const
{
- const size_t ofsMax = m_str.length()*sizeof(wxChar);
+ return m_len;
+}
+
+// ----------------------------------------------------------------------------
+// seek/tell
+// ----------------------------------------------------------------------------
+wxFileOffset wxStringInputStream::OnSysSeek(wxFileOffset ofs, wxSeekMode mode)
+{
switch ( mode )
{
case wxFromStart:
break;
case wxFromEnd:
- ofs += ofsMax;
+ ofs += m_len;
break;
case wxFromCurrent:
break;
default:
- wxFAIL_MSG( _T("invalid seek mode") );
+ wxFAIL_MSG( wxT("invalid seek mode") );
return wxInvalidOffset;
}
- if ( ofs < 0 || wx_static_cast(size_t, ofs) >= ofsMax )
+ if ( ofs < 0 || ofs > static_cast<wxFileOffset>(m_len) )
return wxInvalidOffset;
- m_pos = wx_static_cast(size_t, ofs);
+ // FIXME: this can't be right
+ m_pos = wx_truncate_cast(size_t, ofs);
return ofs;
}
-off_t wxStringInputStream::OnSysTell() const
+wxFileOffset wxStringInputStream::OnSysTell() const
{
- return wx_static_cast(off_t, m_pos);
+ return static_cast<wxFileOffset>(m_pos);
}
// ----------------------------------------------------------------------------
size_t wxStringInputStream::OnSysRead(void *buffer, size_t size)
{
- const size_t sizeMax = m_str.length()*sizeof(wxChar) - m_pos;
+ const size_t sizeMax = m_len - m_pos;
if ( size >= sizeMax )
{
size = sizeMax;
}
- memcpy(buffer, m_str.data() + m_pos, size);
+ memcpy(buffer, m_buf.data() + m_pos, size);
m_pos += size;
return size;
// seek/tell
// ----------------------------------------------------------------------------
-off_t wxStringOutputStream::OnSysTell() const
+wxFileOffset wxStringOutputStream::OnSysTell() const
{
- return wx_static_cast(off_t, m_pos);
+ return static_cast<wxFileOffset>(m_pos);
}
// ----------------------------------------------------------------------------
size_t wxStringOutputStream::OnSysWrite(const void *buffer, size_t size)
{
- // in Unicode mode we might not be able to write the last byte
- size_t len = size / sizeof(wxChar);
+ const char *p = static_cast<const char *>(buffer);
+
+#if wxUSE_UNICODE_WCHAR
+ // the part of the string we have here may be incomplete, i.e. it can stop
+ // in the middle of an UTF-8 character and so converting it would fail; if
+ // this is the case, accumulate the part which we failed to convert until
+ // we get the rest (and also take into account the part which we might have
+ // left unconverted before)
+ const char *src;
+ size_t srcLen;
+ if ( m_unconv.GetDataLen() )
+ {
+ // append the new data to the data remaining since the last time
+ m_unconv.AppendData(p, size);
+ src = m_unconv;
+ srcLen = m_unconv.GetDataLen();
+ }
+ else // no unconverted data left, avoid extra copy
+ {
+ src = p;
+ srcLen = size;
+ }
- const wxChar *p = wx_static_cast(const wxChar *, buffer);
+ size_t wlen;
+ wxWCharBuffer wbuf(m_conv.cMB2WC(src, srcLen, &wlen));
+ if ( wbuf )
+ {
+ // conversion succeeded, clear the unconverted buffer
+ m_unconv = wxMemoryBuffer(0);
- m_str->Append(wxString(p, p + len));
+ m_str->append(wbuf, wlen);
+ }
+ else // conversion failed
+ {
+ // remember unconverted data if there had been none before (otherwise
+ // we've already got it in the buffer)
+ if ( src == p )
+ m_unconv.AppendData(src, srcLen);
+
+ // pretend that we wrote the data anyhow, otherwise the caller would
+ // believe there was an error and this might not be the case, but do
+ // not update m_pos as m_str hasn't changed
+ return size;
+ }
+#else // !wxUSE_UNICODE_WCHAR
+ // no recoding necessary, the data is supposed to already be in UTF-8 (if
+ // supported) or ASCII otherwise
+ m_str->append(p, size);
+#endif // wxUSE_UNICODE_WCHAR/!wxUSE_UNICODE_WCHAR
- // return number of bytes actually written
- len *= sizeof(wxChar);
- m_pos += len;
+ // update position
+ m_pos += size;
- return len;
+ // return number of bytes actually written
+ return size;
}
#endif // wxUSE_STREAMS